144 
which would cause the difference in colour between an 
external and an internal surface, then the formula would he 
q(t + x) = q(t' + x). To find the value of x experimentally, I 
took a solution containing 2400 and sunk a small disc in it 
until the inside and outside colour seemed the same ; for 
the outside the length of column was 22 f 5 ; for the inside 
17 7, this being the mean of eight trials. The difference is 
4-8. With a solution containing 1600, a white surface out- 
side, with length of column 21 % seemed to give the same 
colour as white surface inside with length of column 16*2. 
The difference is 5. 1 also tried to get the value by the 
following combination. A solution was taken, containing 
2400 with disc inside, and compared with a solution contain- 
ing 1600 with a disc outside ; length of column in latter case 
was 21*2, in the former case a column 17*5 seemed to give 
a similar colour. From the formula q(t+x) = q't' the result- 
ing value of x would be 5 2. I also tried the following com- 
bination. Solution containing 2400 and disc outside was 
compared with solution containing 1600 with disc inside — 
the mean of eight trials gave length of column 17 in the 
latter case, equivalent to 21 2 in the former; from the for- 
mula 1600 (17 + $)=2400 X 21*2, the resulting value of x 
is 4’3 — finally we get for the approximate value of x, taking 
the mean of the four determinations, x=4r8 The experi- 
mental determination is not easy, but the value obtained 
gives better results when we use it in the formula ; for in- 
stance, on a former occasion with discs inside, when a 
solution containing 1600 and length of column 21 ‘2 was 
used as a standard, and there was compared with it a solu- 
tion containing 2400, the length of column was 12; from 
the uncorrected formula the result is 2824 ; from the for- 
mula 
^(12 + 4-8) = 1600(21-2 + 4*8) 
the resulting value of q is 2476, which is not far from the 
proper value. When the fluids compared differ much in 
